We are told that Artificial Intelligence is the thing of the future – the universal panacea that will solve humanity’s problems. So far, the results have not been promising. Answers to basic questions have often been incorrect, and in many cases unreal, fictitious legal precedents, false accusations, and libelous photographs seem to be the order of the day. The good news, at least as it relates to our AI mavens, is that no one has determined just how to sue a computer, at least, not yet.

Technology is a tool. Traditionally, it has served either to provide greater opportunities and success for trained workers and educated professionals, or to replace both workers and professionals with machines. In this regard, AI has a serious potential flaw: It assumes that the problem facing humanity is people, particularly their tendency to make mistakes, be corrupt, and/or dishonest. Its proponents believe that AI will solve the problems facing humanity by making people obsolete.

Regrettably, all too often it is those countries and companies most needful of qualified people, that find training and education too costly and demanding. These are the ones who perceive their own workers and staff to be obstacles to development and profit. It is for this reason that these countries and companies remain at the bottom of both the supplier and retailer pile. For example, these countries and their industries look to computerised patternmaking and grading as the way forward, never recognising that unlike the garments produced 50 years ago by professional craftsmen, their hi-tech computer driven solutions make garments that do not fit well. Neither consumer complaints nor returns will sway them from their commitment to the computer driven solution.

Another example would be the hope that AI will provide the necessary information to determine in advance what the consumer wants and which styles will sell best. Once again, these retailers of the future would do well to look back 40 years, when Inditex developed speed to market resulting in trial-orders, fast-turn, and quick-response which turned fashion risk on its head, removing the risk by designing-after-sales.

All of this leads us to the question: What role can Artificial Intelligence play in the global garment industry?

The main advantage of AI is that in its search for answers, it has complete access to the internet universe. The problem is that currently AI lacks discernment:

• The ability to separate the relevant from the irrelevant.

• The ability to determine pedigree-how reliable is the initial data supplier?

• The ability to determine provenance-who has taken the initial data and distorted it to add verisimilitude to themselves and their work?

Nevertheless, I believe that within a limited period of time, AI will be able to deal with many important garment industry problems.

For example, AI can finally solve the garment-fit problem. Our industry made the mistake of replacing body-shape with specifications (specs, for short). In the past, the basic tool of patternmaking was the mannequin. Companies such as Wolf Forms produced three-dimensional bodies custom made to provide the correct shape as defined by the great designers of your choice. They came in all size ranges and product types. In that world of design, size specification was irrelevant. Why base fit on 3, 10, or even 50 measurements, when you had the actual body-shape, which consisted of an almost infinite number of measurements? The mannequin together with the living fit-model ensured that the blouse, dress, pants, coat etc actually fit a human being. The moment specs replaced shape, our industry lost the ability to produce a garment that would fit a person. AI may be able to bring us back to the era when body shape = fit = size. We have the technology to provide the consumer with 3-dimensional sizing – an 8-digit number defining each person’s shape. We have the data defining shape by nationality, age and racial group. We can put the data and technology together to produce a size scale for any consumer, anywhere, at any age, and any race. We can provide the patterns by computer in real time and at almost zero cost.

Another example is drape. Every fabric has a drape. It can be as fine as silk georgette, where the drape can change by colour or as bawdy as 475gms denim. There was a time when professionals could define drape by hand-feel. Today that skillset may exist in haute couture, but the rest of us have real problems. This is an area where AI can play an active role.

The greatest need: Provided it can achieve sufficiently high levels of discernment, as described above, AI may solve the single greatest problem inherent in the internet: Fact checking. Currently, for the most part, we are unable to determine the validity of what we read or hear online. There are two inter-related problems:

First of all, even the most basic and seemingly simplest information becomes complex and nuanced in real life. This is where AI can come to the fore. Consider factory labour rates. For example, what is the monthly wage for a worker in China and how does that compare with the same worker in India? The immediate problem is the need to define location. Providing data for average factory wages on a national level for either China or India is irrelevant. Almost all of China’s garment exports are produced in the eastern provinces, which have the highest wage rates in the country. India faces a similar problem because wage rates and minimum wages differ from one state to another. A second problem is the need to define the working period. Wage rates per month does not provide any meaningful information. Much depends on the number of working days per month as well as the number of working hours per day. At the same time, we must recognise that there is a fundamental difference between wage per month, wage per day and piecework wages. Furthermore, we also have to define wages vs benefits, which might include bonuses, meals, medical care etc. We must differentiate general wages, from factory wages, from sewer factory wages. Finally, we must know the provenance of the data. There is considerable difference between data from the ILO and data from some national garment industry organisation.

Secondly, the most serious problem facing AI is that the data submitted may be intentionally inaccurate. While the internet has provided substantial amounts of useful information, analyses, and data, it has also become the playground for manipulators. We desperately need some tool to differentiate true data, analysis based on fact, knowledgeable and experienced professionals from the deceivers who offer half-truths and factoids. Regrettably, the crooks and conmen tend to be more plausible because plausibility has been their stock-in-trade. We tend to trust the dishonest because they appear to have a better understanding of both our problems and our frustration to find anyone with practical solutions. These same dishonest people offer simple and fast solutions to complex problems which have developed over decades.

We must accept that we are a long way, to the point where AI can take the data and provide true discernment.

Finally, we must return to the beginning. Is the purpose of AI to replace people or to provide a tool for people to do more and better work? This is a fundamental argument that predates AI by approximately 3,000 years. This is the basic question of aesthetics, the difference between talent and craft. There are those who believe that what differentiates people from all other animals is our ability for original thought. On the other side there are those who believe either that a machine can be made capable of original thought or that there is no such thing as original thought. As an unrepentant I must admit that I have little to add on the subject.